Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1/*
2 * net/tipc/msg.c: TIPC message header routines
3 *
4 * Copyright (c) 2000-2006, 2014-2015, Ericsson AB
5 * Copyright (c) 2005, 2010-2011, Wind River Systems
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions are met:
10 *
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the names of the copyright holders nor the names of its
17 * contributors may be used to endorse or promote products derived from
18 * this software without specific prior written permission.
19 *
20 * Alternatively, this software may be distributed under the terms of the
21 * GNU General Public License ("GPL") version 2 as published by the Free
22 * Software Foundation.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34 * POSSIBILITY OF SUCH DAMAGE.
35 */
36
37#include <net/sock.h>
38#include "core.h"
39#include "msg.h"
40#include "addr.h"
41#include "name_table.h"
42
43#define MAX_FORWARD_SIZE 1024
44
45static unsigned int align(unsigned int i)
46{
47 return (i + 3) & ~3u;
48}
49
50/**
51 * tipc_buf_acquire - creates a TIPC message buffer
52 * @size: message size (including TIPC header)
53 *
54 * Returns a new buffer with data pointers set to the specified size.
55 *
56 * NOTE: Headroom is reserved to allow prepending of a data link header.
57 * There may also be unrequested tailroom present at the buffer's end.
58 */
59struct sk_buff *tipc_buf_acquire(u32 size)
60{
61 struct sk_buff *skb;
62 unsigned int buf_size = (BUF_HEADROOM + size + 3) & ~3u;
63
64 skb = alloc_skb_fclone(buf_size, GFP_ATOMIC);
65 if (skb) {
66 skb_reserve(skb, BUF_HEADROOM);
67 skb_put(skb, size);
68 skb->next = NULL;
69 }
70 return skb;
71}
72
73void tipc_msg_init(u32 own_node, struct tipc_msg *m, u32 user, u32 type,
74 u32 hsize, u32 dnode)
75{
76 memset(m, 0, hsize);
77 msg_set_version(m);
78 msg_set_user(m, user);
79 msg_set_hdr_sz(m, hsize);
80 msg_set_size(m, hsize);
81 msg_set_prevnode(m, own_node);
82 msg_set_type(m, type);
83 if (hsize > SHORT_H_SIZE) {
84 msg_set_orignode(m, own_node);
85 msg_set_destnode(m, dnode);
86 }
87}
88
89struct sk_buff *tipc_msg_create(uint user, uint type,
90 uint hdr_sz, uint data_sz, u32 dnode,
91 u32 onode, u32 dport, u32 oport, int errcode)
92{
93 struct tipc_msg *msg;
94 struct sk_buff *buf;
95
96 buf = tipc_buf_acquire(hdr_sz + data_sz);
97 if (unlikely(!buf))
98 return NULL;
99
100 msg = buf_msg(buf);
101 tipc_msg_init(onode, msg, user, type, hdr_sz, dnode);
102 msg_set_size(msg, hdr_sz + data_sz);
103 msg_set_origport(msg, oport);
104 msg_set_destport(msg, dport);
105 msg_set_errcode(msg, errcode);
106 if (hdr_sz > SHORT_H_SIZE) {
107 msg_set_orignode(msg, onode);
108 msg_set_destnode(msg, dnode);
109 }
110 return buf;
111}
112
113/* tipc_buf_append(): Append a buffer to the fragment list of another buffer
114 * @*headbuf: in: NULL for first frag, otherwise value returned from prev call
115 * out: set when successful non-complete reassembly, otherwise NULL
116 * @*buf: in: the buffer to append. Always defined
117 * out: head buf after successful complete reassembly, otherwise NULL
118 * Returns 1 when reassembly complete, otherwise 0
119 */
120int tipc_buf_append(struct sk_buff **headbuf, struct sk_buff **buf)
121{
122 struct sk_buff *head = *headbuf;
123 struct sk_buff *frag = *buf;
124 struct sk_buff *tail = NULL;
125 struct tipc_msg *msg;
126 u32 fragid;
127 int delta;
128 bool headstolen;
129
130 if (!frag)
131 goto err;
132
133 msg = buf_msg(frag);
134 fragid = msg_type(msg);
135 frag->next = NULL;
136 skb_pull(frag, msg_hdr_sz(msg));
137
138 if (fragid == FIRST_FRAGMENT) {
139 if (unlikely(head))
140 goto err;
141 if (unlikely(skb_unclone(frag, GFP_ATOMIC)))
142 goto err;
143 head = *headbuf = frag;
144 *buf = NULL;
145 TIPC_SKB_CB(head)->tail = NULL;
146 if (skb_is_nonlinear(head)) {
147 skb_walk_frags(head, tail) {
148 TIPC_SKB_CB(head)->tail = tail;
149 }
150 } else {
151 skb_frag_list_init(head);
152 }
153 return 0;
154 }
155
156 if (!head)
157 goto err;
158
159 if (skb_try_coalesce(head, frag, &headstolen, &delta)) {
160 kfree_skb_partial(frag, headstolen);
161 } else {
162 tail = TIPC_SKB_CB(head)->tail;
163 if (!skb_has_frag_list(head))
164 skb_shinfo(head)->frag_list = frag;
165 else
166 tail->next = frag;
167 head->truesize += frag->truesize;
168 head->data_len += frag->len;
169 head->len += frag->len;
170 TIPC_SKB_CB(head)->tail = frag;
171 }
172
173 if (fragid == LAST_FRAGMENT) {
174 TIPC_SKB_CB(head)->validated = false;
175 if (unlikely(!tipc_msg_validate(head)))
176 goto err;
177 *buf = head;
178 TIPC_SKB_CB(head)->tail = NULL;
179 *headbuf = NULL;
180 return 1;
181 }
182 *buf = NULL;
183 return 0;
184err:
185 pr_warn_ratelimited("Unable to build fragment list\n");
186 kfree_skb(*buf);
187 kfree_skb(*headbuf);
188 *buf = *headbuf = NULL;
189 return 0;
190}
191
192/* tipc_msg_validate - validate basic format of received message
193 *
194 * This routine ensures a TIPC message has an acceptable header, and at least
195 * as much data as the header indicates it should. The routine also ensures
196 * that the entire message header is stored in the main fragment of the message
197 * buffer, to simplify future access to message header fields.
198 *
199 * Note: Having extra info present in the message header or data areas is OK.
200 * TIPC will ignore the excess, under the assumption that it is optional info
201 * introduced by a later release of the protocol.
202 */
203bool tipc_msg_validate(struct sk_buff *skb)
204{
205 struct tipc_msg *msg;
206 int msz, hsz;
207
208 if (unlikely(TIPC_SKB_CB(skb)->validated))
209 return true;
210 if (unlikely(!pskb_may_pull(skb, MIN_H_SIZE)))
211 return false;
212
213 hsz = msg_hdr_sz(buf_msg(skb));
214 if (unlikely(hsz < MIN_H_SIZE) || (hsz > MAX_H_SIZE))
215 return false;
216 if (unlikely(!pskb_may_pull(skb, hsz)))
217 return false;
218
219 msg = buf_msg(skb);
220 if (unlikely(msg_version(msg) != TIPC_VERSION))
221 return false;
222
223 msz = msg_size(msg);
224 if (unlikely(msz < hsz))
225 return false;
226 if (unlikely((msz - hsz) > TIPC_MAX_USER_MSG_SIZE))
227 return false;
228 if (unlikely(skb->len < msz))
229 return false;
230
231 TIPC_SKB_CB(skb)->validated = true;
232 return true;
233}
234
235/**
236 * tipc_msg_build - create buffer chain containing specified header and data
237 * @mhdr: Message header, to be prepended to data
238 * @m: User message
239 * @dsz: Total length of user data
240 * @pktmax: Max packet size that can be used
241 * @list: Buffer or chain of buffers to be returned to caller
242 *
243 * Returns message data size or errno: -ENOMEM, -EFAULT
244 */
245int tipc_msg_build(struct tipc_msg *mhdr, struct msghdr *m,
246 int offset, int dsz, int pktmax, struct sk_buff_head *list)
247{
248 int mhsz = msg_hdr_sz(mhdr);
249 int msz = mhsz + dsz;
250 int pktno = 1;
251 int pktsz;
252 int pktrem = pktmax;
253 int drem = dsz;
254 struct tipc_msg pkthdr;
255 struct sk_buff *skb;
256 char *pktpos;
257 int rc;
258
259 msg_set_size(mhdr, msz);
260
261 /* No fragmentation needed? */
262 if (likely(msz <= pktmax)) {
263 skb = tipc_buf_acquire(msz);
264 if (unlikely(!skb))
265 return -ENOMEM;
266 skb_orphan(skb);
267 __skb_queue_tail(list, skb);
268 skb_copy_to_linear_data(skb, mhdr, mhsz);
269 pktpos = skb->data + mhsz;
270 if (copy_from_iter(pktpos, dsz, &m->msg_iter) == dsz)
271 return dsz;
272 rc = -EFAULT;
273 goto error;
274 }
275
276 /* Prepare reusable fragment header */
277 tipc_msg_init(msg_prevnode(mhdr), &pkthdr, MSG_FRAGMENTER,
278 FIRST_FRAGMENT, INT_H_SIZE, msg_destnode(mhdr));
279 msg_set_size(&pkthdr, pktmax);
280 msg_set_fragm_no(&pkthdr, pktno);
281 msg_set_importance(&pkthdr, msg_importance(mhdr));
282
283 /* Prepare first fragment */
284 skb = tipc_buf_acquire(pktmax);
285 if (!skb)
286 return -ENOMEM;
287 skb_orphan(skb);
288 __skb_queue_tail(list, skb);
289 pktpos = skb->data;
290 skb_copy_to_linear_data(skb, &pkthdr, INT_H_SIZE);
291 pktpos += INT_H_SIZE;
292 pktrem -= INT_H_SIZE;
293 skb_copy_to_linear_data_offset(skb, INT_H_SIZE, mhdr, mhsz);
294 pktpos += mhsz;
295 pktrem -= mhsz;
296
297 do {
298 if (drem < pktrem)
299 pktrem = drem;
300
301 if (copy_from_iter(pktpos, pktrem, &m->msg_iter) != pktrem) {
302 rc = -EFAULT;
303 goto error;
304 }
305 drem -= pktrem;
306
307 if (!drem)
308 break;
309
310 /* Prepare new fragment: */
311 if (drem < (pktmax - INT_H_SIZE))
312 pktsz = drem + INT_H_SIZE;
313 else
314 pktsz = pktmax;
315 skb = tipc_buf_acquire(pktsz);
316 if (!skb) {
317 rc = -ENOMEM;
318 goto error;
319 }
320 skb_orphan(skb);
321 __skb_queue_tail(list, skb);
322 msg_set_type(&pkthdr, FRAGMENT);
323 msg_set_size(&pkthdr, pktsz);
324 msg_set_fragm_no(&pkthdr, ++pktno);
325 skb_copy_to_linear_data(skb, &pkthdr, INT_H_SIZE);
326 pktpos = skb->data + INT_H_SIZE;
327 pktrem = pktsz - INT_H_SIZE;
328
329 } while (1);
330 msg_set_type(buf_msg(skb), LAST_FRAGMENT);
331 return dsz;
332error:
333 __skb_queue_purge(list);
334 __skb_queue_head_init(list);
335 return rc;
336}
337
338/**
339 * tipc_msg_bundle(): Append contents of a buffer to tail of an existing one
340 * @skb: the buffer to append to ("bundle")
341 * @msg: message to be appended
342 * @mtu: max allowable size for the bundle buffer
343 * Consumes buffer if successful
344 * Returns true if bundling could be performed, otherwise false
345 */
346bool tipc_msg_bundle(struct sk_buff *skb, struct tipc_msg *msg, u32 mtu)
347{
348 struct tipc_msg *bmsg;
349 unsigned int bsz;
350 unsigned int msz = msg_size(msg);
351 u32 start, pad;
352 u32 max = mtu - INT_H_SIZE;
353
354 if (likely(msg_user(msg) == MSG_FRAGMENTER))
355 return false;
356 if (!skb)
357 return false;
358 bmsg = buf_msg(skb);
359 bsz = msg_size(bmsg);
360 start = align(bsz);
361 pad = start - bsz;
362
363 if (unlikely(msg_user(msg) == TUNNEL_PROTOCOL))
364 return false;
365 if (unlikely(msg_user(msg) == BCAST_PROTOCOL))
366 return false;
367 if (unlikely(msg_user(bmsg) != MSG_BUNDLER))
368 return false;
369 if (unlikely(skb_tailroom(skb) < (pad + msz)))
370 return false;
371 if (unlikely(max < (start + msz)))
372 return false;
373 if ((msg_importance(msg) < TIPC_SYSTEM_IMPORTANCE) &&
374 (msg_importance(bmsg) == TIPC_SYSTEM_IMPORTANCE))
375 return false;
376
377 skb_put(skb, pad + msz);
378 skb_copy_to_linear_data_offset(skb, start, msg, msz);
379 msg_set_size(bmsg, start + msz);
380 msg_set_msgcnt(bmsg, msg_msgcnt(bmsg) + 1);
381 return true;
382}
383
384/**
385 * tipc_msg_extract(): extract bundled inner packet from buffer
386 * @skb: buffer to be extracted from.
387 * @iskb: extracted inner buffer, to be returned
388 * @pos: position in outer message of msg to be extracted.
389 * Returns position of next msg
390 * Consumes outer buffer when last packet extracted
391 * Returns true when when there is an extracted buffer, otherwise false
392 */
393bool tipc_msg_extract(struct sk_buff *skb, struct sk_buff **iskb, int *pos)
394{
395 struct tipc_msg *msg;
396 int imsz, offset;
397
398 *iskb = NULL;
399 if (unlikely(skb_linearize(skb)))
400 goto none;
401
402 msg = buf_msg(skb);
403 offset = msg_hdr_sz(msg) + *pos;
404 if (unlikely(offset > (msg_size(msg) - MIN_H_SIZE)))
405 goto none;
406
407 *iskb = skb_clone(skb, GFP_ATOMIC);
408 if (unlikely(!*iskb))
409 goto none;
410 skb_pull(*iskb, offset);
411 imsz = msg_size(buf_msg(*iskb));
412 skb_trim(*iskb, imsz);
413 if (unlikely(!tipc_msg_validate(*iskb)))
414 goto none;
415 *pos += align(imsz);
416 return true;
417none:
418 kfree_skb(skb);
419 kfree_skb(*iskb);
420 *iskb = NULL;
421 return false;
422}
423
424/**
425 * tipc_msg_make_bundle(): Create bundle buf and append message to its tail
426 * @list: the buffer chain, where head is the buffer to replace/append
427 * @skb: buffer to be created, appended to and returned in case of success
428 * @msg: message to be appended
429 * @mtu: max allowable size for the bundle buffer, inclusive header
430 * @dnode: destination node for message. (Not always present in header)
431 * Returns true if success, otherwise false
432 */
433bool tipc_msg_make_bundle(struct sk_buff **skb, struct tipc_msg *msg,
434 u32 mtu, u32 dnode)
435{
436 struct sk_buff *_skb;
437 struct tipc_msg *bmsg;
438 u32 msz = msg_size(msg);
439 u32 max = mtu - INT_H_SIZE;
440
441 if (msg_user(msg) == MSG_FRAGMENTER)
442 return false;
443 if (msg_user(msg) == TUNNEL_PROTOCOL)
444 return false;
445 if (msg_user(msg) == BCAST_PROTOCOL)
446 return false;
447 if (msz > (max / 2))
448 return false;
449
450 _skb = tipc_buf_acquire(max);
451 if (!_skb)
452 return false;
453
454 skb_trim(_skb, INT_H_SIZE);
455 bmsg = buf_msg(_skb);
456 tipc_msg_init(msg_prevnode(msg), bmsg, MSG_BUNDLER, 0,
457 INT_H_SIZE, dnode);
458 if (msg_isdata(msg))
459 msg_set_importance(bmsg, TIPC_CRITICAL_IMPORTANCE);
460 else
461 msg_set_importance(bmsg, TIPC_SYSTEM_IMPORTANCE);
462 msg_set_seqno(bmsg, msg_seqno(msg));
463 msg_set_ack(bmsg, msg_ack(msg));
464 msg_set_bcast_ack(bmsg, msg_bcast_ack(msg));
465 tipc_msg_bundle(_skb, msg, mtu);
466 *skb = _skb;
467 return true;
468}
469
470/**
471 * tipc_msg_reverse(): swap source and destination addresses and add error code
472 * @own_node: originating node id for reversed message
473 * @skb: buffer containing message to be reversed; may be replaced.
474 * @err: error code to be set in message, if any
475 * Consumes buffer at failure
476 * Returns true if success, otherwise false
477 */
478bool tipc_msg_reverse(u32 own_node, struct sk_buff **skb, int err)
479{
480 struct sk_buff *_skb = *skb;
481 struct tipc_msg *hdr = buf_msg(_skb);
482 struct tipc_msg ohdr;
483 int dlen = min_t(uint, msg_data_sz(hdr), MAX_FORWARD_SIZE);
484
485 if (skb_linearize(_skb))
486 goto exit;
487 hdr = buf_msg(_skb);
488 if (msg_dest_droppable(hdr))
489 goto exit;
490 if (msg_errcode(hdr))
491 goto exit;
492
493 /* Take a copy of original header before altering message */
494 memcpy(&ohdr, hdr, msg_hdr_sz(hdr));
495
496 /* Never return SHORT header; expand by replacing buffer if necessary */
497 if (msg_short(hdr)) {
498 *skb = tipc_buf_acquire(BASIC_H_SIZE + dlen);
499 if (!*skb)
500 goto exit;
501 memcpy((*skb)->data + BASIC_H_SIZE, msg_data(hdr), dlen);
502 kfree_skb(_skb);
503 _skb = *skb;
504 hdr = buf_msg(_skb);
505 memcpy(hdr, &ohdr, BASIC_H_SIZE);
506 msg_set_hdr_sz(hdr, BASIC_H_SIZE);
507 }
508
509 /* Now reverse the concerned fields */
510 msg_set_errcode(hdr, err);
511 msg_set_origport(hdr, msg_destport(&ohdr));
512 msg_set_destport(hdr, msg_origport(&ohdr));
513 msg_set_destnode(hdr, msg_prevnode(&ohdr));
514 msg_set_prevnode(hdr, own_node);
515 msg_set_orignode(hdr, own_node);
516 msg_set_size(hdr, msg_hdr_sz(hdr) + dlen);
517 skb_trim(_skb, msg_size(hdr));
518 skb_orphan(_skb);
519 return true;
520exit:
521 kfree_skb(_skb);
522 *skb = NULL;
523 return false;
524}
525
526/**
527 * tipc_msg_lookup_dest(): try to find new destination for named message
528 * @skb: the buffer containing the message.
529 * @err: error code to be used by caller if lookup fails
530 * Does not consume buffer
531 * Returns true if a destination is found, false otherwise
532 */
533bool tipc_msg_lookup_dest(struct net *net, struct sk_buff *skb, int *err)
534{
535 struct tipc_msg *msg = buf_msg(skb);
536 u32 dport, dnode;
537 u32 onode = tipc_own_addr(net);
538
539 if (!msg_isdata(msg))
540 return false;
541 if (!msg_named(msg))
542 return false;
543 if (msg_errcode(msg))
544 return false;
545 *err = -TIPC_ERR_NO_NAME;
546 if (skb_linearize(skb))
547 return false;
548 msg = buf_msg(skb);
549 if (msg_reroute_cnt(msg))
550 return false;
551 dnode = addr_domain(net, msg_lookup_scope(msg));
552 dport = tipc_nametbl_translate(net, msg_nametype(msg),
553 msg_nameinst(msg), &dnode);
554 if (!dport)
555 return false;
556 msg_incr_reroute_cnt(msg);
557 if (dnode != onode)
558 msg_set_prevnode(msg, onode);
559 msg_set_destnode(msg, dnode);
560 msg_set_destport(msg, dport);
561 *err = TIPC_OK;
562 return true;
563}
564
565/* tipc_msg_reassemble() - clone a buffer chain of fragments and
566 * reassemble the clones into one message
567 */
568struct sk_buff *tipc_msg_reassemble(struct sk_buff_head *list)
569{
570 struct sk_buff *skb;
571 struct sk_buff *frag = NULL;
572 struct sk_buff *head = NULL;
573 int hdr_sz;
574
575 /* Copy header if single buffer */
576 if (skb_queue_len(list) == 1) {
577 skb = skb_peek(list);
578 hdr_sz = skb_headroom(skb) + msg_hdr_sz(buf_msg(skb));
579 return __pskb_copy(skb, hdr_sz, GFP_ATOMIC);
580 }
581
582 /* Clone all fragments and reassemble */
583 skb_queue_walk(list, skb) {
584 frag = skb_clone(skb, GFP_ATOMIC);
585 if (!frag)
586 goto error;
587 frag->next = NULL;
588 if (tipc_buf_append(&head, &frag))
589 break;
590 if (!head)
591 goto error;
592 }
593 return frag;
594error:
595 pr_warn("Failed do clone local mcast rcv buffer\n");
596 kfree_skb(head);
597 return NULL;
598}